JPH04372469A - Rear wheel steering device for vehicle - Google Patents

Abstract

PURPOSE:To set the rear wheel steering angle coping with the driver's demand and faculty according to the cornering situation. CONSTITUTION:When a driver does not agree with the rear wheel steering angle instruction value outputted by a controller C, the driver himself inputs the correction signal into a teacher signal setting circuit 10. In the teacher signal setting circuit 10 which receives the correction signal corrects the multiplication factor and sets the teacher signal to be inputted into a neural net 8, and outputs the teacher signal into the neural net. In the neural net which receives the teacher signal corrects the own weight so that the difference between the teacher signal and the output value calculated by the neural net may become less.

Description

[Detailed description of the invention]

0001

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a rear wheel steering system that can set a rear wheel steering angle according to a driver's preference and ability in accordance with cornering conditions.

0002

[Prior Art] As a device of this kind, Japanese Patent Application Laid-Open No. 63-28
A rear wheel steering device according to No. 7676 has been known for some time. This conventional device includes a steering angle sensor and a vehicle speed sensor, and a controller processes signals from these sensors to determine the rear wheel steering angle.

0003

[Problems to be Solved by the Invention] With the conventional device as described above, the driver cannot set or correct the rear wheel steering angle according to his/her preference. Therefore, the meaningful parameters for determining the rear wheel steering angle are:
From a safety perspective, everything was set with beginners in mind. However, experienced drivers are not satisfied with the beginner-level settings, and there is a problem in that they are not able to fully demonstrate the ability of the rear wheel steering device. An object of the present invention is to provide a rear wheel steering device that allows the rear wheel steering angle to be set according to the driver's preference and ability.

0004

[Means for Solving the Problems] The present invention provides a controller that outputs a rear wheel steering angle instruction signal in response to the input of vehicle motion parameters such as a steering wheel angle, steering wheel torque, or vehicle speed, and a controller that outputs a rear wheel steering angle instruction signal in response to input of vehicle motion parameters such as a steering wheel angle, steering wheel torque, or vehicle speed. a neural network that learns the movement situation, a switch connected to this neural network, and a teacher signal that is connected to the neural network via the switch and sets the correction signal input by the driver as the teacher signal of the neural network. A setting circuit, a multiplier that is connected to the neural network via the switch and multiplies the output signal of the controller by the output signal of the neural network, and a rear wheel actuator such as a power cylinder according to the output signal of this multiplier. It is characterized in that it is equipped with an actuator driver that drives the.

[0005]

[Operation] Since the rear wheel steering device of the present invention is constructed as described above, if the driver is not satisfied with the rear wheel steering angle instruction value outputted by the controller, the driver himself/herself can send a correction signal to the driver. Input to signal setting circuit. The teacher signal setting circuit that receives the correction signal corrects the multiplier, sets the teacher signal to be input to the neural network, and outputs it to the neural network. The neural network that receives this teacher signal modifies its own weights so that the difference between the teacher signal and the output value calculated by the neural network becomes smaller.

[0006]

[Effects of the Invention] According to the rear wheel steering device of the present invention, the driver can set the rear wheel steering angle himself when cornering, so it can accommodate the preferences and abilities of everyone from beginners to veterans.

[0008]

[Embodiment] A steering wheel angle sensor 1, a steering wheel torque sensor 2, and a vehicle speed sensor 3 are connected to the controller C of the illustrated embodiment, and the rear wheel steering angle is calculated by the controller C using the detection signals of these sensors as parameters. , print it.

A multiplier 4 is connected to the controller C as described above, and an actuator driver 6 for driving a rear wheel actuator 5 such as a power cylinder is further connected to the multiplier 4.

[0010] Each of the above-mentioned sensors 1 to 3 is also a sampler.
It is connected to a neural network 8 via 7. This neural network 8 is connected to a multiplier 4 and a teacher signal setting circuit 10 via a switch 9. When a correction signal determined by the driver's will is input to the teacher signal setting circuit 10, the current multiplier is corrected and set as a teacher signal for the neural network 8. The neural network 8 that has received this modified teacher signal modifies its own weights so that the difference between the teacher signal and the output value calculated by the neural network 8 becomes smaller. Incidentally, the output signal of the neural network 8 is always inputted to the teacher signal setting circuit 10 in order to grasp the multiplier before correction.

The specific configuration of the neural network 8 as described above is shown in FIG. Further, the learning starts from the time when the driver inputs a correction signal to the teacher signal setting circuit 10. At this time, data n steps past the time when the correction signal was input is input to the neural network 8. For example, if the sampling interval is 0.5 seconds and n=10, then data up to 5 seconds ago will be input. This series of input data indicates one state of vehicle motion, and under this state, the output value of the neural network 8 and the correction signal from the teacher signal setting circuit 10 are compared, and the difference is determined. Modify the weights of the neural net 8 so that they become smaller. In this way, the neural net 8 learns what the multiplier value is according to the driver's preference in one vehicle motion state.

Next, the operation of this embodiment will be explained. During normal driving, switch 9 is switched to the right side, which is the opposite side as shown in the drawing. Then, when the vehicle travels, signals indicating the motion status of the vehicle are input to the controller C from each of the sensors 1 to 3. The controller C calculates a rear wheel steering calculation value based on each of the above signals and outputs it to the multiplier 4.

The output signals of the sensors 1 to 3 are also input to a neural network 8 via a sampler 7. The neural network 8 outputs the learned multiplier value according to the output signal from each sensor, and transmits it to the multiplier 4. Therefore, the multiplier 4 multiplies the rear wheel steering angle calculation value from the controller C by the multiplier value from the neural network 8, outputs a rear wheel steering angle command value, and adjusts the actuator driver according to the output signal. Drive 6.

As described above, the rear wheel actuator 5 operates according to the rear wheel steering angle instruction value output from the multiplier 4, but if the driver is not satisfied with the steering situation, the driver can adjust the steering angle according to his own preference. A correction signal corresponding to the above is input to the teacher signal setting circuit 10. At this time, the neural network 8 has inputted data n steps past from that point.

Furthermore, the teacher signal setting circuit 10 modifies the multipliers so far and sets them as the teacher signal for the neural network 8. Neural network 8 that received this teacher signal
corrects its own weights so that the difference between the corrected teacher signal and the output value calculated by the neural network until now becomes small. Therefore, from then on, multiplier 4
, the modified multiplier from the neural net 8 will be input.

The neural network 8 outputs the multiplier modified as described above, and if the driver is still not satisfied when cornering is performed based on the multiplier, the teacher signal setting circuit 10 is operated again to determine the current value. Modify the multiplier up to and set its teacher signal. By repeating such operations, the neural network 8 corrects its weights by itself and exhibits a learning effect.

As described above, according to the rear wheel steering device of this embodiment, the rear wheel steering angle can be set to suit the preference and ability of the driver depending on the cornering situation, so it is suitable for drivers ranging from beginners to experienced drivers. It can handle all of the following.

[Brief explanation of drawings]

FIG. 1 is a circuit diagram.

FIG. 2 is a configuration diagram of a neural network.

[Explanation of symbols]

1 Handle angle sensor 2 Handle torque sensor 3 Vehicle speed sensor C Controller 4 Multiplier 5 Rear wheel actuator 6 Actuator driver 8 Neural network 9 Switch 10 Teacher signal setting circuit

Claims (1)

[Claims] [Claim 1] A controller that outputs a rear wheel steering angle instruction signal in response to input of vehicle motion parameters such as steering wheel angle, steering wheel torque, or vehicle speed, and that learns the motion status by inputting these vehicle motion parameters. A neural net, a switch connected to this neural net, a teacher signal setting circuit that is connected to the neural net via the switch and sets the corrected signal input by the driver as a teacher signal of the neural network, and a multiplier that multiplies the output signal of the controller by the output signal of the neural network, and an actuator driver that drives a rear wheel actuator such as a power cylinder according to the output signal of this multiplier. Rear wheel steering system for vehicles.